> How do LLMs do it? They don't learn by writing a lot they learn by reading a lot.
Literally they learn by both, training process is filled with "exercises" for the weights to be shaped by, not just "seeing" tokens :) At least the typical LLM that outputs tokens have to actually "practice" doing so.
Literally they learn by both, training process is filled with "exercises" for the weights to be shaped by, not just "seeing" tokens :) At least the typical LLM that outputs tokens have to actually "practice" doing so.